2023-07-02-fan23b.md

abstract	openreview	title	layout	series	publisher	issn	id	month	tex_title	firstpage	lastpage	page	order	cycles	bibtex_author	author	date	address	container-title	volume	genre	issued	pdf	extras
Data for pretraining machine learning models often consists of collections of heterogeneous datasets. Although training on their union is reasonable in agnostic settings, it might be suboptimal when the target domain —where the model will ultimately be used— is known in advance. In that case, one would ideally pretrain only on the dataset(s) most similar to the target one. Instead of limiting this choice to those datasets already present in the pretraining collection, here we explore extending this search to all datasets that can be synthesized as ‘combinations’ of them. We define such combinations as multi-dataset interpolations, formalized through the notion of generalized geodesics from optimal transport (OT) theory. We compute these geodesics using a recent notion of distance between labeled datasets, and derive alternative interpolation schemes based on it: using either barycentric projections or optimal transport maps, the latter computed using recent neural OT methods. These methods are scalable, efficient, and —notably— can be used to interpolate even between datasets with distinct and unrelated label sets. Through various experiments in transfer learning in computer vision, we demonstrate this is a promising new approach for targeted on-demand dataset synthesis.	ZXF4YgdsfJT	Generating Synthetic Datasets by Interpolating along Generalized Geodesics	inproceedings	Proceedings of Machine Learning Research	PMLR	2640-3498	fan23b	0	Generating Synthetic Datasets by Interpolating along Generalized Geodesics	571	581	571-581	571	false	Fan, Jiaojiao and Alvarez-Melis, David	given family Jiaojiao Fan given family David Alvarez-Melis	2023-07-02		Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence	216	inproceedings	date-parts 2023 7 2	https://proceedings.mlr.press/v216/fan23b/fan23b.pdf	label link Supplementary PDF https://proceedings.mlr.press/v216/fan23b/fan23b-supp.pdf